Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed Network
The current work proposes a novel super-resolution convolutional transposed network (SRCTN) deep learning architecture for downscaling daily climatic variables. The algorithm was established based on a super-resolution convolutional neural network with transposed convolutions. This study designed sy...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-01-01
|
| Series: | Water |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4441/16/2/335 |
| _version_ | 1797342386618105856 |
|---|---|
| author | Yong Liu Xiaohui Yan Wenying Du Tianqi Zhang Xiaopeng Bai Ruichuan Nan |
| author_facet | Yong Liu Xiaohui Yan Wenying Du Tianqi Zhang Xiaopeng Bai Ruichuan Nan |
| author_sort | Yong Liu |
| collection | DOAJ |
| description | The current work proposes a novel super-resolution convolutional transposed network (SRCTN) deep learning architecture for downscaling daily climatic variables. The algorithm was established based on a super-resolution convolutional neural network with transposed convolutions. This study designed synthetic experiments to downscale daily reference evapotranspiration (ET<sub>0</sub>) data, which are a key indicator for climate change, from low resolutions (2°, 1°, and 0.5°) to a fine resolution (0.25°). The entire time period was divided into two major parts, i.e., training–validation (80%) and test periods (20%), and the training–validation period was further divided into training (80%) and validation (20%) parts. In the comparison of the downscaling performance between the SRCTN and Q-M models, the root-mean-squared error (RMSE) values indicated the accuracy of the models. For the SRCTN model, the RMSE values were reported for different scaling ratios: 0.239 for a ratio of 8, 0.077 for a ratio of 4, and 0.015 for a ratio of 2. In contrast, the RMSE values for the Q-M method were 0.334, 0.208, and 0.109 for scaling ratios of 8, 4, and 2, respectively. Notably, the RMSE values in the SRCTN model were consistently lower than those in the Q-M method across all scaling ratios, suggesting that the SRCTN model exhibited better downscaling performance in this evaluation. The results exhibited that the SRCTN method could reproduce the spatiotemporal distributions and extremes for the testing period very well. The trained SRCTN model in one study area performed remarkably well in a different area via transfer learning without re-training or calibration, and it outperformed the classic downscaling approach. The good performance of the SRCTN algorithm can be primarily attributed to the incorporation of transposed convolutions, which can be partially seen as trainable upsampling operations. Therefore, the proposed SRCTN method is a promising candidate tool for downscaling daily ET<sub>0</sub> and can potentially be employed to conduct downscaling operations for other variables. |
| first_indexed | 2024-03-08T10:32:33Z |
| format | Article |
| id | doaj.art-800963741b27420bba6909ed2f1b4cd2 |
| institution | Directory Open Access Journal |
| issn | 2073-4441 |
| language | English |
| last_indexed | 2024-03-08T10:32:33Z |
| publishDate | 2024-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Water |
| spelling | doaj.art-800963741b27420bba6909ed2f1b4cd22024-01-26T18:52:13ZengMDPI AGWater2073-44412024-01-0116233510.3390/w16020335Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed NetworkYong Liu0Xiaohui Yan1Wenying Du2Tianqi Zhang3Xiaopeng Bai4Ruichuan Nan5Artificial Intelligence Key Laboratory of Sichuan Province, Yibin 643000, ChinaArtificial Intelligence Key Laboratory of Sichuan Province, Yibin 643000, ChinaNational Engineering Research Center for Geographic Information System, China University of Geosciences, Wuhan 430074, ChinaDepartment of Water Resources Engineering, Dalian University of Technology, Dalian 116024, ChinaDepartment of Water Resources Engineering, Dalian University of Technology, Dalian 116024, ChinaDepartment of Water Resources Engineering, Dalian University of Technology, Dalian 116024, ChinaThe current work proposes a novel super-resolution convolutional transposed network (SRCTN) deep learning architecture for downscaling daily climatic variables. The algorithm was established based on a super-resolution convolutional neural network with transposed convolutions. This study designed synthetic experiments to downscale daily reference evapotranspiration (ET<sub>0</sub>) data, which are a key indicator for climate change, from low resolutions (2°, 1°, and 0.5°) to a fine resolution (0.25°). The entire time period was divided into two major parts, i.e., training–validation (80%) and test periods (20%), and the training–validation period was further divided into training (80%) and validation (20%) parts. In the comparison of the downscaling performance between the SRCTN and Q-M models, the root-mean-squared error (RMSE) values indicated the accuracy of the models. For the SRCTN model, the RMSE values were reported for different scaling ratios: 0.239 for a ratio of 8, 0.077 for a ratio of 4, and 0.015 for a ratio of 2. In contrast, the RMSE values for the Q-M method were 0.334, 0.208, and 0.109 for scaling ratios of 8, 4, and 2, respectively. Notably, the RMSE values in the SRCTN model were consistently lower than those in the Q-M method across all scaling ratios, suggesting that the SRCTN model exhibited better downscaling performance in this evaluation. The results exhibited that the SRCTN method could reproduce the spatiotemporal distributions and extremes for the testing period very well. The trained SRCTN model in one study area performed remarkably well in a different area via transfer learning without re-training or calibration, and it outperformed the classic downscaling approach. The good performance of the SRCTN algorithm can be primarily attributed to the incorporation of transposed convolutions, which can be partially seen as trainable upsampling operations. Therefore, the proposed SRCTN method is a promising candidate tool for downscaling daily ET<sub>0</sub> and can potentially be employed to conduct downscaling operations for other variables.https://www.mdpi.com/2073-4441/16/2/335evapotranspirationsuper resolutionconvolutional neural networkdownscaling |
| spellingShingle | Yong Liu Xiaohui Yan Wenying Du Tianqi Zhang Xiaopeng Bai Ruichuan Nan Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed Network Water evapotranspiration super resolution convolutional neural network downscaling |
| title | Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed Network |
| title_full | Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed Network |
| title_fullStr | Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed Network |
| title_full_unstemmed | Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed Network |
| title_short | Downscaling Daily Reference Evapotranspiration Using a Super-Resolution Convolutional Transposed Network |
| title_sort | downscaling daily reference evapotranspiration using a super resolution convolutional transposed network |
| topic | evapotranspiration super resolution convolutional neural network downscaling |
| url | https://www.mdpi.com/2073-4441/16/2/335 |
| work_keys_str_mv | AT yongliu downscalingdailyreferenceevapotranspirationusingasuperresolutionconvolutionaltransposednetwork AT xiaohuiyan downscalingdailyreferenceevapotranspirationusingasuperresolutionconvolutionaltransposednetwork AT wenyingdu downscalingdailyreferenceevapotranspirationusingasuperresolutionconvolutionaltransposednetwork AT tianqizhang downscalingdailyreferenceevapotranspirationusingasuperresolutionconvolutionaltransposednetwork AT xiaopengbai downscalingdailyreferenceevapotranspirationusingasuperresolutionconvolutionaltransposednetwork AT ruichuannan downscalingdailyreferenceevapotranspirationusingasuperresolutionconvolutionaltransposednetwork |